Why Don’t Cat Models Work or Do They? The Role of Models: A View From Both Sides CAS Presentation May 8, 2006 Maria Kovas
Overview Role of Modeling during the Quiet Years Original Role of Modeling Change in Expectations Return to the basics
Post Event Loss Evaluation During the Quiet Natural Catastrophe Years Leading Role in: Reinsurance Pricing Underwriting Primary Pricing Business Planning Post Event Loss Evaluation
Original Role: Diagnostic Device Distribution of Exposure Identification of High Density Areas Review of Adjacency of Risk in Key Peril Areas Consideration of Redistribution of Exposure
Identification of Drivers of Risk Review of PML in Modeled for Modeled Perils Application of Secondary Peril information Modeled Un-Modeled
Consideration of Data Quality and the Impact on the Diagnostic Process Inclusion of additional sublimits and deductibles Identification of more building characteristics Cost Benefit analysis of modifying systems to update the data
Metrics Determination of Business/Underwriting Goals Review the current position of the organization Chart the differences and devise options to achieve Goals Periodic review and updates of movement in the process
Underwriting Cycle
Catastrophic Response It is not possible to plan for a Catastrophe It is possible to evaluate event curves Event drills Deployment of claims teams Setting management expectations
Factors Contributing to the Change in Expectations: Peril Modeling vs. Total losses Second Generation Modelers The Quiet Period Data Issues Updates to methodologies
Peril Modeling vs. Total losses Models are based on specific perils Review of actual events to calibrate model outputs Quiet period limited the number events for calibration Actual losses are based on all affected perils
Second Generation Modelers First Generation-Professional insurance/broker/reinsurers-recruited to use the models because of aptitude or interest Second Generation-Professional modelers recruited for technical aptitude and skill in running the models
The Quiet Period Modeling became the sole focus of loss estimates Comfort level because losses were within a tolerable variance
Data Issues Models capable of refinement with additional data points Data capture may be not updated to match the capabilities of the models Valuation issues New financial structures
Updates to methodologies New interpretation of historic data Updates to the science used in the models Changes to the methodology Re-calibration of models and vulnerabilities
Re-establishing the Role of Modeling Model data as Currency Return to the basics Ensemble Reporting Setting expectations
Model data as Currency: Definition of the data coin A data formatted in such a manner to be shared among industry business interests Standard evaluation of the data coin determine the confidence in the data Confidence in the data leads to rate considerations
Return to Basics Underwriting Cycle Review Reporting provided to management Recreate event drills Refresh training Update data capture and interface
Ensemble Reporting Create Multiple Views of Exposure Model what is available Identify and communicate gaps Evaluate Secondary Perils Use Mapping
Communicate Anticipate questions Consider key issues in the firm Review the diagnosis Determine if the remedies are cost effective